Longitudinal effects of aging on 18F-FDG distribution in cognitively normal elderly individuals

Previous studies of aging effects on fluorine-18-labeled fluorodeoxyglucose (18F-FDG) distribution have employed cross-sectional designs. We examined aging effects on 18F-FDG distribution using both cross-sectional and longitudinal assessments. We obtained two 18F-FDG positron emission tomography scans at two different time points from 107 cognitively normal elderly participants. The participants’ mean ages at baseline and second scans were 67.9 and 75.7, respectively. The follow-up period ranged from 4 to 11 years with a mean of 7.8 years. The voxel-wise analysis revealed significant clusters in which 18F-FDG uptake was decreased between baseline and second scans (p < 0.05, family-wise error corrected) in the anterior cingulate cortex (ACC), posterior cingulate cortex/precuneus (PCC/PC), and lateral parietal cortex (LPC). The cross-sectional analysis of 18F-FDG uptake and age showed significant correlations in the ACC (p = 0.016) but not the PCC/PC (p = 0.240) at baseline, and in the ACC (p = 0.004) and PCC/PC (p = 0.002) at the second scan. The results of longitudinal assessments suggested that 18F-FDG uptake in the ACC, PCC/PC, and LPC decreased with advancing age in cognitively normal elderly individuals, and those of the cross-sectional assessments suggested that the trajectories of age-associated 18F-FDG decreases differed between the ACC and PCC/PC.

PET acquisition and image processing. The radioligand, 18 F-FDG, was synthesized using a PET synthesizer (Sumitomo Heavy Industries, Ltd., Tokyo, Japan). The radiochemical purity of 18 F-FDG was greater than 95%. The PET scanning was performed at the Tokyo Metropolitan Institute of Gerontology using a SET-2400W scanner (Shimadzu Corporation, Kyoto, Japan) in the three-dimensional mode. The in-plane and axial resolutions of the full width at half maximum were 4.4 mm and 6.5 mm, respectively. The transmission data were acquired with a rotating 68 Ga/ 68 Ge rod source for measured attenuation correction. The emission data were acquired for 6 min starting 45 min after an intravenous bolus injection of approximately 150 MBq of 18 F-FDG. Sixty-three-slice images with a voxel size of 2.054 × 2.054 × 3.125 mm 3 and matrix size of 128 × 128 were obtained. The data were reconstructed after correcting for decay, attenuation, and scatter.
The images were processed using the FMRIB Software Library (version 5.0.4; Oxford University, Oxford, UK). All 18 F-FDG images were nonlinearly transformed into the Montreal Neurological Institute (MNI) standard space from the native space using an in-house developed 18 F-FDG template. The transformed images were globally normalized using a cortical mask developed in-house as a reference region (Fig. 1A,B). The mean value within the masked voxels was set to 1, and the normalized images were smoothed with an isotropic Gaussian kernel with a sigma of 4 mm for the subsequent voxel-wise and volume-of-interest (VOI) analyses. We also specified sex and MMSE scores as nuisance covariates to exclude any of their effects. Correction for multiple comparisons was applied to the whole-brain voxel-wise analysis using a family-wise error (FWE) approach. The FWE-corrected threshold was set at p < 0.05. A VOI analysis was then performed to test the longitudinal association between 18 F-FDG uptake and age, and the influence of ApoE ε4 genotype on 18 F-FDG distribution. First, VOIs were manually placed on the ACC and posterior cingulate cortex/precuneus (PCC/PC) in the MNI standard space, where voxel-wise analysis detected highly significant clusters (Fig. 1C). The VOI volumes were 992 mm 3 (124 voxels) and 784 mm 3 (98 voxels) for the ACC and PCC/PC, respectively. These VOIs were superimposed on normalized 18 F-FDG images in the MNI standard space, and the values for the VOIs were extracted. For each of the ACC and PCC/PC VOIs, we also calculated the annual rate of reduction in normalized 18 F-FDG uptake (%) as follows: 100 × [(VOI value at baseline) − (VOI value at second PET)]/(VOI value at baseline)/(time interval from baseline to second PET). The annual rates in the ACC and PCC/PC were compared between groups 2 and 3 (i.e., presence or absence of ApoE ε4 genotype) using Student's t-tests. P values less than 0.05 were considered statistically significant.
Cross-sectional analysis. In order to integrate the results of the longitudinal analysis, a cross-sectional analysis was performed on the data from the baseline and second PET scans and VOIs described above. We performed a correlational analysis of the relationship between normalized 18 F-FDG uptake and age in the ACC and PCC/PC. P values less than 0.05 were considered statistically significant. All statistical analyses were conducted using SPSS Statistics version 22 (IBM, Armonk, NY). Data Availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Results
The exploratory whole-brain voxel-wise analysis using a two-way repeated-measures analysis of variance revealed significant effects of aging on 18 F-FDG distribution at p < 0.05, FWE-corrected (Fig. 2). The clusters that extended through the ACC, PCC/PC, and lateral parietal cortex (LPC) were highly significant, suggesting that 18 F-FDG uptake decreased in these areas with advancing age. The cluster size in the ACC and PCC/PC was 10737 voxels. The significant clusters in the left and right LPC were 2070 voxels and 568 voxels, respectively. However, the effects of the ApoE ε4 genotype on 18 F-FDG distribution were not significant in both the positive and negative contrasts at p < 0.05, FWE-corrected.
The results of the longitudinal VOI analysis of the data from the ACC and PCC/PC are displayed in Figs 3 and 4. For each participant, the changes in 18 F-FDG uptake from the baseline scan to the second PET scan are plotted in Fig. 3. We visually confirmed that these results were consistent with the results shown in Fig. 2, which suggested that 18 F-FDG uptake in the ACC and PCC/PC tended to decrease with advancing age. The annual rates of reduction in 18 F-FDG uptake were plotted in the ACC and PCC/PC for each group (Fig. 4). The rates in the AC and PCC/PC in the group-1 individuals were 0.58 ± 0.55% and 0.57 ± 0.53% (mean ± SD), respectively. The corresponding rates were 0.57 ± 0.55% and 0.54 ± 0.53%, respectively, in the group-2 individuals, and 0.67 ± 0.55% and 0.71 ± 0.50%, respectively, in the group-3 individuals. The annual rates did not significantly differ between the group-2 and -3 individuals in the ACC (p = 0.49) and PCC/PC (p = 0.20).
The results of the cross-sectional analysis of the VOIs in the ACC and PCC/PC are displayed in Fig. 5. The relationships between 18 F-FDG uptake and age at the baseline and second PET scans were plotted in the ACC and PCC/PC. In the baseline scans (Fig. 5, left), a significant correlation was found in the ACC (r = 0.233, p = 0.016) but not the PCC/PC (r = 0.114, p = 0.240). In contrast, significant correlations were found in the ACC (r = 0.274, p = 0.004) and PCC/PC (r = 0.303, p = 0.002) in the second PET scans (Fig. 5, right).

Discussion
The primary objective of this study was to investigate the longitudinal effects of aging on 18 F-FDG distribution in the brains of cognitively normal elderly individuals. We showed that 18 F-FDG uptake in the ACC, PCC/PC, and LPC was longitudinally decreased with higher statistical significance in elderly individuals. The follow-up period ranged from 4 to 11 years with a mean of 7.8 years. To the best of our knowledge, this is the first longitudinal study investigating aging effects on 18 F-FDG distribution in a large number of elderly individuals with a longer follow-up period. Recent studies addressing aging effects on 18 F-FDG distribution [13][14][15] have focused primarily on the progression of AD, with fewer participants and shorter follow-up periods. In a study by Shokouhi and colleagues with a mean follow up on 4.3 years, 18 F-FDG distribution was fairly stable over time in 24 healthy elderly participants 15 . Similarly, no regional changes in 18 F-FDG distribution were observed in a study by Ossenkoppele and colleagues with 11 healthy elderly participants and mean follow-up period of 2.5 years 14 . These may indicate that a follow-up period of 2-4 years is too short to investigate aging effects on 18 F-FDG distribution in healthy  elderly individuals. On the other hand, recent cross-sectional 18 F-FDG PET studies in young to elderly individuals have consistently shown that the most prominent age-associated decrease in 18 F-FDG uptake was in anterior regions of the brain, including the ACC, and that the LPC is one of the areas in which 18 F-FDG uptake decreases with advancing aging [5][6][7][8][9] . Recent studies have also reported that part of the PCC/PC is an area that exhibits an age-associated decrease in 18 F-FDG uptake, with relatively low statistical significance [6][7][8] . Therefore, our results were in line with those of previous cross-sectional 18 F-FDG PET studies.
Compared to the results of the cross-sectional studies that investigated the association between 18 F-FDG uptake and aging in young to elderly individuals, the present longitudinal study detected a highly significant association in the PCC/PC in addition to the ACC (Fig. 2). These findings may be explained as follows. The magnitude of the 18 F-FDG decrease in younger to elderly individuals is much larger in the ACC than it is in the PCC/PC, while the magnitude of the 18 F-FDG decrease in the ACC is comparable to the decrease in the PCC/ PC over old age. In fact, the differences in glucose metabolism between young and elderly individuals tend to be greater in anterior regions of the brain 2 . Fjell and colleagues recently investigated the effects of aging on regional brain volume in a cross-sectional study of 1,100 healthy adults (18-94 years), and they identified 3 basic types of age-associated trajectories: (1) a linear reduction, (2) stability followed by a decline, and (3) a steep nonlinear decline 21 . They showed that the volumes in the amygdala, nucleus accumbens, and striatum decreased linearly in an age-associated manner from young adulthood. After a period of relative stability during middle age, the volume of the hippocampus continuously decreased over old age.  Interestingly, as part of the temporoparietal networks, the PCC/PC is functionally connected with the hippocampus [22][23][24] , and, as part of the frontostriatal networks, the ACC is functionally connected with the ventral striatum, including the nucleus accumbens 23,25 . In these networks, changes in regional volume with advancing age 21 may remotely affect regional metabolic changes. Thus, a linear volumetric reduction in the nucleus accumbens may cause linear metabolic reductions in the ACC, and stability followed by an accelerated volumetric decline in the hippocampus may similarly cause metabolic reductions in the PCC/PC. These speculations might be supported by the results shown in Fig. 5 that 18 F-FDG uptake decreased continuously in the ACC, while 18 F-FDG uptake in the PCC/PC was stable until around age 60-70 but then started to decease in an age-associated manner. Based on previous findings and our results, we propose a model of the effects of aging on 18 F-FDG uptake in the ACC and PCC/PC (Fig. 6), although future studies are needed to validate this speculative model.
One of the limitations of the present study was the absence of amyloid-β (Aβ) PET imaging in the cognitively normal elderly participants. Because the PCC/PC and LPC are areas that exhibit decreased 18 F-FDG uptake in patients with AD 26,27 , one cannot deny that some of the participants in this study might have already been in a preclinical stage of AD. Previous Aβ PET imaging studies have found that about 10-30% of cognitively normal elderly participants present significant Aβ deposition [28][29][30][31][32] . The ApoE ε4 genotype is associated with higher Aβ retention 29 , and possibly causes AD-like hypometabolism in cognitively normal elderly participants 33 . Therefore, in order to exclude as many participants as possible who might be in a preclinical stage of AD, we set up group 2, which consisted of participants without the ApoE ε4 genotype. This allowed us to assess the influence of ApoE ε4 genotype on 18 F-FDG distribution. The results of the voxel-wise analysis and comparison of the annual rates of reduction in 18 F-FDG uptake (Fig. 4) suggest that the longitudinal 18 F-FDG reduction in the ACC and PCC/ PC occurs regardless of ApoE genotype. However, because our sample size was unequal between groups 2 and 3, further investigations are needed to validate our results.
Aβ deposition reportedly starts to develop more than 20 years before the onset of cognitive decline, which suggests that the preclinical stage of AD can last over than 20 years 13,32,34 . Additionally, 18 F-FDG uptake in the PC, which shows the earliest metabolic decline in the brain, reportedly starts to decrease about 14 years before AD onset 13 . Some longitudinal studies have found that cognitively normal individuals with Aβ deposition are at higher risk for future cognitive decline than those without Aβ deposition 35,36 . However, such a long duration of the preclinical stage of AD suggests that the preclinical stage of AD might be part of the normal aging process in some elderly individuals 23 . The differences between normally aging individuals and cognitively normal elderly individuals with Aβ deposition or other latent diseases are still unclear, which makes it difficult to define normal aging. In other words, identifying elderly individuals who might be in a preclinical stage of undetected dementia is extremely difficult. Additional longitudinal follow-up investigations of the participants in this study will be needed to further elucidate this issue.
The PCC/PC and LPC are the main components of the default mode network (DMN), which plays an important role in regulating complex cognition and behavior [37][38][39] . Its function is affected by aging 23,40 . Thus, the DMN is vulnerable to aging, and functional disruption in the DMN has been reported to be associated with cognitive decline in aging 41 . Considering the relationship between the DMN and aging, our findings of the longitudinal decreases in 18 F-FDG in the PCC/PC and LPC might partially reflect aging-induced reductions in the functional connectivity of the DMN.
Another limitation of this study was that we could not perform the MRI-based partial volume correction (PVC) due to the absence of adequate MRI data. Because the mean interval between PET scans was 7.8 years, the volume of the brain likely decreased over the interval of this study. Therefore, the lack of PVC method might affect 18 F-FDG measurements due to aging related atrophy. This limitation needs to be resolved in subsequent studies.
Both males and females show age-associated 18 F-FDG decline, predominantly in anterior regions of the brain 6,8,12 . However, sex differences in aging effects on 18 F-FDG distribution are controversial. Some researchers suggest that sex differences are non-significant or minimal; however, others suggest that males show age-associated 18 F-FDG decline in wider cortical areas than females 5,12 . Furthermore, sex differences in 18 F-FDG uptake may depend on age; the difference may be smaller in elderly compared to middle aged people 6 . We could Figure 6. Schematic model of the age-associated trajectories. The blue and red lines represent the ageassociated decreases in 18 F-FDG uptake in the ACC and PCC/PC, respectively. In the ACC (blue), 18 F-FDG uptake decreased continuously over time. In the PCC/PC, 18 F-FDG uptake was relatively stable before age 60-70 and then started to decease in an age-associated manner. 18  not address sex differences in this study because the number of people based on sex in our sample was unequal. Further investigations are also needed to address this issue.

Conclusions
The results of the present longitudinal study showed that 18 F-FDG uptake in the ACC, PCC/PC, and LPC of cognitively normal elderly individuals decreased highly significantly with advancing age and that these findings remained regardless of ApoE ε4 genotype. The cross-sectional results showed that 18 F-FDG uptake in the ACC decreased continuously with age, while 18 F-FDG uptake in the PCC/PC was stable until around age 60-70 at which point it began to decease. These findings suggest that the trajectories of age-associated decreases in 18 F-FDG uptake differ between the ACC and PCC/PC. However, because of the limitations of this study, such as the absence of Aβ assessments, additional longitudinal follow-up investigations in this sample are needed to confirm that our findings are due to aging effects.